The disclosed inventions relate generally to conducting electronic auctions, and in particular to business-to-business bidding auctions for industrial purchasers.
Traditional Procurement Models
Procurement of supplies has traditionally involved high transaction costs, especially information search costs. The introduction of electronic commerce has introduced new methods of procurement that lower some of the transaction costs associated with procurement. Online procurement, or business-to-business electronic commerce, matches buyers and suppliers so that transactions can take place electronically. There are three models for online procurement: catalog, buyer-bidding auction, and seller-bidding auction.
The "catalog" model of online procurement was the first to be developed. The first electronic catalogs were developed by suppliers to help customers obtain information about products and order supplies electronically. These first electronic catalogs were single-source; i.e. they only allowed customers to obtain information and products from that supplier.
However, customers were not satisfied with being "locked in" to one supplier--they wanted to be able to compare a number of competing products to be sure of getting the product features they wanted, at the best price. So suppliers with single-source electronic catalogs started to include competitors' products on their systems. An example of this is American's SABRE system, which includes offerings from competing suppliers (airlines), thereby further reducing information search costs. By offering competing products, the electronic catalog that offers competitor's products becomes an "electronic market".
Many of these systems are biased towards the supplier offering the electronic market. Procurement costs can be further lowered with an unbiased electronic market that promotes competition.
For standard products and services, the need to have an unbiased market has been met for many industries by third party "market makers." For example, Inventory Locator Services has compiled a database that lists all airplane parts suppliers that have a certain item in stock. Buyers dial into the database to get information on the parts they need. Here, it is a third party, Inventory Locator Service, not a supplier, creating the unbiased electronic market.
The electronic catalog model of electronic commerce involves one buyer and one seller at a time. When many buyers compete for the right to buy from one seller, a buyer-bidding auction model is created. A noteworthy example of the buyer-bidding auction model is that operated by PriceLine.com and described in U.S. Pat. No. 5,794,207 issued to Walker et al. In this system, potential buyers compete for airline tickets by submitting a bid for an airline ticket on the PriceLine website, and airlines can choose to accept a bid, thereby committing the buyer to buy the ticket.
The catalog and buyer-bidding auction types of electronic markets do not work in some situations however. If the required product is custom, it is not possible for suppliers to publish a set price for a catalog market. Likewise, it is not possible for buyers to identify the product they want to bid on in a buyer-bidding auction. There are fewer suppliers and no standard product and pricing information available for the buyer of custom industrial products. Traditionally, when a company requires a custom industrial product, procurement is made by a buyer for the company who searches for a supplier and acquires price quotes from a potential supplier for the needed custom product. The search is slow and somewhat random because it usually relies heavily on personal relationships. The costs associated with locating vendors, comparing their products, negotiating, and paperwork are a big factor in the make-or-buy decision. The cost of switching suppliers is very large, which means that the quoted price is probably not the lowest fair price and that it is hard for a new supplier to enter the market.
Therefore, buyers wanted to use auctions to save money. The assignee of the present application developed a system wherein sellers downwardly bid against one another to achieve the lowest market price in a supplier-bidding auction.
Supplier-Bidding Auction
In a supplier-bidding auction, bid prices start high and move downward in reverse-auction format as bidders interact to establish a closing price. The auction marketplace is one-sided, i.e. one buyer and many potential suppliers. Typically, the products being purchased are components or materials. "Components" typically mean fabricated tangible pieces or parts that become part of assemblies of durable products. Example components include gears, bearings, appliance shelves or door handles. "Materials" typically mean bulk quantities of raw materials that are further transformed into product. Example materials include corn syrup or sheet steel.
Industrial buyers do not typically purchase one component at a time. Rather, they purchase whole families of similar components. At times, components are strongly related to one another. As an example, a buyer might purchase a given plastic knob in two different colors, or might purchase a nameplate in four different languages. These parts are so similar that by definition they must be purchased from the same supplier--all of the knobs are made using the same mold. These items are therefore grouped into a single lot. Bidders in industrial auctions must provide unit price quotes for all line items in a lot.
Auction Process
The process for a supplier-bidding auction as conducted by the assignee of the present application is described below with reference to FIGS. 1 and 2. FIG. 1 illustrates the functional elements and entities in an supplier-bidding auction, while FIG. 2 is a process diagram that identifies the tasks performed by each of the involved entities.
The supplier-bidding auction model requires that the bidding product or service be defined by the buyer (identified as Buyer 10 in FIG. 1). An auction coordinator (Coordinator 20 in FIG. 1) works with buyers to prepare for and conduct an auction and to define the potentially new supply relationships resulting from the auction.
As shown in FIG. 2, in the Initial Contact phase 102 of the auction process, the coordinator contacts the buyer, and the buyer provides data to the coordinator. The coordinator prepares a specification 50 for each desired product or part 52. Once the product 52 is defined, potential suppliers for the product are identified. The coordinator 20 and buyer 10 work together to compile this list of potential suppliers from suppliers already known to the buyer as well as suppliers recommended by the coordinator.
The buyer makes a decision regarding which potential suppliers will receive invitations to the upcoming Auction. Suppliers that accept Auction invitations are then sent notices regarding the upcoming Auction, as well as client software to install in preparation of participating the Auction.
In the RFQ phase 104, coordinator 20 works with the buyer 10 to prepare a Request for Quotation ("RFQ") 54. The coordinator collects and maintains the RFQ data provided by buyer 10, and then publishes the RFQ, and manages the published RFQ. The RFQ includes specifications 50 for all of the parts 52 covered by the RFQ. In the RFQ 54, buyer 10 aggregates similar part or commodity line items into job "lots." These lots allow suppliers 30 to bid on that portion of the business for which they are best suited.
During the auction 56, bids 58 will be taken against individual lots (and their constituent parts 52) within RFQ 54. While bidders must submit actual unit prices for all line items, the competition in an Auction is based on the aggregate value bid for lots. The aggregate value bid for a lot depends upon the level and mix of line item bids and the quantity for each line item. Therefore, bidders submit bids at the line item level, but compete on the lot level.
In the Auction Administration phase 106, coordinator 20 coordinates the Auction and administers the Auction setup and preparation. The coordinator sends a RFQ to each participating supplier, and assists participating suppliers with preparation for the Auction.
In the Auction phase 108, suppliers 30 submit bids 58 on the lots and monitor the progress of the bidding by the participating suppliers 30. The coordinator assists, observes, and administers the Auction.
When the bidding period is over, the auction enters the Auction Results Administration phase 110. In this phase, coordinator 20 analyzes and administers the Auction results, which are viewed by buyer 10. The buyer begins to conduct final qualification of the low bidding supplier(s). The buyer may retain the right not to award business to a low bidding supplier based on final qualification results or other business concerns.
In the ensuing Contract Administration phase 112, the coordinator facilitates settlements 62 awarded by the buyer 10 to suppliers 30. Contracts 52 are then drawn up between buyer 10 and suppliers 30.
Communications and Software
The Auction is conducted electronically between potential suppliers 30 at their respective remote sites and the coordinator 20 at its site. As shown in FIGS. 3 and 4, information is conveyed between the coordinator 20 and the suppliers 30 via a communications medium such as a network service provider 40 accessed by the participants through, for example, dial-up telephone connections using modems, or direct network connections. A computer software application is used to manage the Auction. The software application has two components: a client component 31 and a server component 23. The client component 31 operates on a computer at the site of each of the potential suppliers 30. The client component is used by suppliers 30 to make bids during the Auction. The bids are sent via the network service provider 40 to the site of the coordinator, where it is received by the server component 23 of the software application. The client component includes software used to make a connection through telephone lines or the Internet to the server component. Bids are submitted over this connection and updates are sent to connected bidders.
Bids can only be submitted using the client component of the application--this ensures that buyers do not circumvent the bidding process, and that only invited suppliers participate in the bidding. Typically, bidders can see their bids and bids placed by other suppliers for each lot on the client component. When a bidder submits a bid, that bid is sent to the server component and evaluated to determine whether the bid is from an authorized bidder, and whether the bid has exceeded a pre-determined maximum acceptable price. Bids placed by a supplier are broadcast to all connected bidders thereby enabling every participating bidder to see quickly the change in market conditions and begin planning their competitive responses.
Conduct of an Auction
The conduct of an Auction will now be described in conjunction with the operation of the software application. The Auction is conducted on a specified date, and commences at a specified time. Bidding on each of the lots of products involved is scheduled to begin simultaneously at the start time for the Auction. Each lot is assigned a scheduled closing time after which further bids by potential suppliers submitted via the client application will not be accepted by the server application. The closing times for the lots are staggered so that they are not coterminous.
Associated with each lot at any given time in the progress of the Auction is a bidding status. The possible bidding statuses are illustrated in FIG. 5. The status initially assigned to each lot, before the scheduled start time of the Auction, is "Available." This status indicates that the lot will be available for bidding in the Auction. In the normal sequence of an Auction, the next bidding status is "Open," which indicates that the Auction is underway and that bids can be submitted for the lot. There are two possible bidding statuses to which a lot with an "Open" status can change: "Overtime" and "Closed." Overtime indicates an extension of time to allow bidding to continue after the scheduled closing time for bidding on the lot. If bidding is still active at the end of a first Overtime period of predetermined duration, the server application allows a second Overtime, and so on, until bidding has closed. "Closed" indicates that the server application will no longer accept bids on the lot. A lot's status changes from Overtime only to Closed.
Information regarding the Auction that can be displayed by the client application is illustrated in FIGS. 6A-6D at selected times during the conduct of an Auction. FIG. 6A illustrates lot information provided at the start of an Auction. The lot closing times are shown for each of the lots (01-08) as 10:30:00 AM, etc. The lots are identified by name (e.g. "PP-Gas assist" which stands for "Gas Assist Polypropylene Parts" for lot 01). The indicated status for each lot is "OPEN." The "Market Bid" column indicates the current lowest or best bid for the lot. The current time (10:00:11 AM) is shown in the upper right corner of the display.
The presented information changes during the course of bidding. For purposes of illustration in this example, a series of bids for lot 01 is shown in FIG. 8. Selected bids are identified by an "Event" code (A, B, etc.) in the first column. The bidder's identity is shown in the second column. The time at which the bid was submitted and the amount of the bid are shown in the third and fourth columns. Finally, the best bid in existence at the time of each subsequent bid is shown in the last column.
The changes in status of lot 01 are also illustrated in FIG. 7 for selected times and corresponding bidding events during the Auction. FIG. 7A shows a time line for lot 01, with the bid event letters corresponding to the bids in FIG. 8.
This Auction employs a decision rule to trigger overtime that can be stated: "when a low bid is submitted during a first time interval t before the scheduled close, reschedule the close to occur later by one time interval t." Thus, for a time interval t of one minute, a scheduled closing time of 10:30 is extended to 10:31 if overtime is triggered.
In this example, Bid A is received at 10:26:49. This bid has no effect on the status or on the scheduled closing time of lot 01, because it does not arrive within one minute of the scheduled closing time of 10:30.
When bid B is received, the status of lot 01 immediately changes to Overtime, because bid B is a low bid and is received at 10:29:06, within one minute of 10:30. The scheduled closing time is therefore delayed until 10:31, which is one increment t (one minute) after the original closing time of 10:30. This additional increment is available for bidders to consider whether to submit a bid in response to bid B.
When bid C is received, the status of the lot remains overtime, but because it was a new low bid and was received at 10:30:03, within one increment t of the then-scheduled closing time of 10:31, the scheduled closing time is further delayed by one increment t to 10:32.
When bid D is received at 10:30:45, there is no effect on the status of the scheduled closing time, because although the bid is a new low bid ($371,373) it is not received within one minute of the then-scheduled closing time of 10:32.
Although Bid E is received within one minute of the scheduled closing time of 10:32, it is not a new low bid (i.e. is greater than 371,373) and therefore no additional time is added. Lot 01 therefore closes at 10:32.
FIG. 6B shows the status of the Auction at 10:27. Lot 01 is shown as "Open," with a current Market Bid (best current bid) of $374,586. This reflects the status after bidding several bids have been received. FIG. 6C illustrates the status of the Auction at 10:30 AM, after bids B and C. In bid B, Bidder15 submitted another best bid ($373,063), which initiated a one-minute overtime period, extending the closing time for lot 1 to 10:31. In bid C, Bidder7 submitted another best bid ($372,500), which initiated another one-minute overtime period, extending the closing time for lot 1 to 10:32. Bid D, submitted at 10:30:45 was another best bid ($371,373), but was not submitted within one minute of the 10:32 closing time. Bid E, although submitted at 10:31:45, within one minute of closing, was not a new low bid and therefore did not extend the closing time for Lot 1. Lot 1 therefore closed at 10:32, with a Market Bid of $371,373, as shown in FIG. 6D (which shows the status of the Auction at 10:32:05).
Bidding Dynamics
Suppliers prepare their price quotes in light of a number of factors. These factors include raw material prices, the design of existing dies or fixtures, the dimensional tolerance required of the component, the amount of engineering support the purchaser desires, the speed with which this particular buyer pays invoices, and the distance the product must be shipped.
Supplier-specific factors also affect the price quotes. Capacity availability, desirability of this particular buyer as a customer, desired levels of profit, and desire to diversify into other markets can all affect the price the seller is willing to accept to supply the needed product. Market-specific factors that are not predictable during the preparation of quotations but that are evident during the auction can also be important in determining, for example, how aggressively other participants may bid.
Because business-to-business auctions are conducted for important custom components, low bidders may still be "passed over" if other bidders demonstrate non-price advantages.
Sometimes auctions involve parts that this purchaser has procured before, and are possibly being made currently by one or more suppliers. These would be termed "existing parts." When a part is currently being made by a supplier, that supplier would be termed the "incumbent supplier." In an auction situation, the incumbent supplier is placed in a position of having to defend its contract with the purchaser.
Incumbent suppliers are expected to behave differently than outsiders. An incumbent, for example, knows that the buyer's switching costs favor the incumbent even at a price premium to the market. Because the buyer may pass over low bidders incumbent suppliers can take advantage of their incumbent status.
Certain human factors must also be considered when conducting business-to-business auctions for industrial purchasing. If not considered, these human factors can interfere with achieving desired outcomes. Bidders must be comfortable with the auction software. Bidders often speak English as a second language, or not at all, making it desirable to provide interpreters.
Bidders must often respond to multi-million-dollar decisions in a few seconds. The fast response required creates cognitive limits--a bidder cannot realistically focus on more than one decision at a time. Many bidders are under some sort of emotional stress when participating, due to the change involved. In some cases, incumbent bidders are literally "fighting for their lives" in situations where losing the contract in question literally means losing their business.
Problems with Prior Auction Process
The prior auction process described above has been found to produce suboptimal results for buyers in light of the market dynamics issues identified above in some circumstances. The problems include: a) multiple lot closing time collisions; b) premature lot closings; c) difficult and inflexible bidding constraints due to lot/line item structure; d) possible prejudice to bidders resulting from technical disruptions; and e) possible prejudice to bidders resulting from submission of erroneous bids.
The first problem is collision of closing times for multiple lots. As described above, the initially scheduled closing times for multiple lots are staggered, so that the lots close at different times, with 10-20 minute intervals between lots. This allows suppliers who would like to bid on multiple lots the opportunity to do so, without having to bid on each lot at the same time. By spacing the closing times for each lot, a supplier knows that while bidding on one lot, the next lot in the Auction will not close. This staggered closing is one way to work around cognitive limits--each lot is sequenced so that bidders can pay attention to one lot at a time.
Overtime delays on an early lot can reduce the time interval before the scheduled closing time of a subsequent lot. In fact, overtime delays have at times overrun the scheduled closing time of a subsequent lot. This situation begins to tax cognitive limits, and bidding opportunities on the subsequent lots are often missed. Although this problem might be resolved by spacing market closing times at a substantial distance apart, experience has shown that even this approach would not be sufficient. For example, in one Auction with 2 lots scheduled with 20 minutes between closing times, actual bidding on the first lot continued for 5 hours and 31 minutes after scheduled close.
The second problem is premature closing of bidding on lots. Just as in an in-person auction, bidding activity tends to increase close to the scheduled closing time. Like the "going, going, gone" auction concept, it is possible to achieve a better auction price if the auction is allowed to continue if bids are still being made. As described above, this concept in implemented through the use of "Overtime," by which the closing time of a specific lot is automatically extended based on the flow of bids into the Auction. Overtime prevents bidders from hanging back and submitting last minute bids in an attempt to prevent competitive reaction.
In the prior system, if a valid, low bid is received in a specified time interval before a lot's scheduled closing time, then the closing time is amended (delayed) to give other bidders more time to react to the late-arriving bid. A bid for second place, something an incumbent supplier might be expected to do, would not trigger an overtime. Unlike in-person auctions, industrial auctions need to allow second place bids. Thus, it has been found that this is an overly simplistic model, which may still be cutting off bidding too soon. A low bidder needs a chance to react to a second place bid, but it cannot if that second place bid does not trigger overtime. Therefore, it would desirable to be able to trigger Overtime, or extend Overtime, on the basis of more complex bid scenarios.
It has also been found that it would be preferable to have more flexibility in the operation of overtime. Overtime is currently implemented with fixed offsets between lot closing times, with fixed trigger time frames (the period before scheduled closing or current Overtime ending), and fixed extension periods for Overtime. In industrial markets, bidding events involve commodities of varying complexity and component packages of different sizes. Accordingly, bidders may need more or less overtime to respond to a bid. The amount of overtime may need to be customized for specific bidding events or for individual lots within an Auction to obtain the optimal market dynamic. It is therefore desirable to provide more flexibility in Overtime.
Another difficulty encountered by bidders in the prior system is that bids needed to be made at the line item level. However, the auction takes place at the lot level, where all of the line item quotes are added up to one sum. This dual structure results from the nature of the industrial market. Price quotations may be built from many cost elements that are added up to form price quotes for individual items or groups of items. Some cost elements or items within a market may be negotiable and others may be fixed depending on factors specific to each bidder. For example, a fabricator may not have control of raw material costs. The competition for a lot can thus involve many independent and dependent factors. Bids are placed in real-time and often in rapid successions. This requires bidders to quickly adjust price quotations for an entire lot, even though this lot may be comprised of hundreds of individual items or cost elements. Initial online auctions only allowed line item price adjustments, and it was difficult to fine-tune the mix of bids. A bidder would have to individually adjust line items, while the software calculated new total lot price bids. This was frequently too time-consuming to keep pace with the auction.
With the time constraints on bidding, bidders wanted the ability to be able to rapidly adjust the lot price without specifically changing individual line items. Therefore, "pro rata" bid adjustment was developed. With pro rata bid adjustment, bidders could change the total lot price quote, and the software would apportion pro rata the change across individual line items in the lot. However, a pro rata apportionment may result in individual items being priced at levels that are uneconomic for the bidder. For some items the bidders may wish to set decision rules for adjusting the unit price quotes at the line item level. For example, bidders may wish to lock-in a preset floor or ceiling on particular line items within a low. These decision rules, or "locks" cause the pro rata adjustments to be applied selectively instead of universally.
Another problem that can arise either in regular bidding or in Overtime is addressing technical disruptions. Real-time technical or operational disruptions in the communications network, software or hardware during the course of a bidding event may prevent a bidder from fully participating. Disruptions may arise in the online network or due to the equipment used by an individual bidder.
In addition, market events or imperfections may disrupt bidding activity and require communication with bidders before the auction can continue. For example, in one auction bidding for an auction lot commenced and it became clear that some of the bidders were including tooling costs and some were not. The correct assumption was to exclude tooling costs. In this example, the bidders needed to be contacted and informed of the correction before the auction could proceed with all bidders on an equal footing.
Other external factors may disrupt the operation of the auction or participation by bidders. In another example, a snow storm prevented many bidders from getting to work on time for the opening of the auction. This was not discovered until many bidders failed to commence bidding.
With a business-to-business transaction, it is typically commercially unacceptable for any bidder to be denied full participation. However, where disruptions arise in the course of a bidding event, the bidding activity and positions of other bidders cannot be prejudiced. Accordingly, market closing times may need to be suspended pending the resolution of a disruption experienced by one or more bidders. This is especially true where it is not clear how long it will take to resolve the problems encountered, including whether it will be possible to resolve all problems prior to the close of an auction lot.
Another problem that occurred in the previous system was the submission of erroneous bids. Bidding errors can happen due to the nature of the online auction. The pace or intensity of the bidding activity can exceed cognitive limits of bidders. In an effort to keep up, bidders enter incorrect quote amounts. Bidders are frequently interested in bidding on multiple lots. In the course of monitoring or switching between lots, the bidder erroneously enters a bid intended for Lot A into Lot B.
In the industrial market, the stakes can be quite large, with contracts awarded for millions of dollars and for contracts that can last for several years. Therefore, the economic damage to suppliers that would be incurred by honoring an erroneous bid is substantial.
In an online industrial auction, an incorrect bid can upset the bidding behavior of other bidders creating inequity for all participants in the market. All bidders in the marketplace view market conditions based on bids placed by other bidders, and respond. They do so relying on the implicit assumption that all bids are valid. If one of the bids is made in error, and the other bidders proceed to bid in response, the integrity of the auction is damaged. This can result in sub-optimal results for both buyers and suppliers.